On a Framework for more Automatic Design and Research in Sheet Metal Forming: A Case Study on Applied Automatic Optimisation of Process Simulations
Translated title
On a Framework for more Automatic Design and Research in Sheet Metal Forming
Authors
Bøjesen, Dennis ; Klose, Claes Rubin
Term
4. term
Education
Publication year
2018
Pages
239
Abstract
Dette speciale adresserer de stigende krav til design og procesindstillinger i pladeformgivning, som følger af bl.a. udbredelsen af højstyrkestål, ved at udvikle et udgangspunkt for mere automatisk optimering af processimuleringer. Der foreslås et rammeværk med fem faser – Formål, Proces, Modellering, Optimering samt Resultater og Diskussion – hvor numeriske modeller paramétreres og kobles til en automatisk optimeringssløjfe. Optimeringsproblemerne formuleres som ikke-lineær mindste kvadraters problemer og løses med en gradientbaseret algoritme i en trust-region-ramme. Rammeværket anvendes i seks cases: en industriel fin-pladeproces med kvalitetsproblemer, en ny totrinsmetode til at opnå små radier (hydromekanisk dybtræk og efterfølgende bulging-pressing), samme metode på en kompleks geometri, en hydraulisk ekspanderet rør–tubesheet-samling, et Airbus A380 dørkarmhjørne med tilbagefjedring og brudstendenser, samt en undersøgelse af en metode til at reducere restspændinger i en dybtrukken kop via egenformanalyse. På tværs af casene demonstreres, at den automatiske optimering kan udpege relevante design- og procesparametre og, hvor det er relevant, forbedre resultater som værktøjsgeometri og tilbagefjedring; i dørkarmcasen gav en optimeret form eksempelvis forbedret tilbagefjedring. Afslutningsvis evalueres rammeværket, og der skitseres retningslinjer for, hvordan sådanne optimeringer kan tilrettelægges i praksis.
This thesis addresses the growing need for robust design and process settings in sheet metal forming, driven by the wider use of high-strength steels, by developing a starting point for more automatic optimization of process simulations. It proposes a five-phase framework—Purpose, Process, Modelling, Optimization, and Results & Discussion—in which numerical models are parameterized and integrated into an automatic optimization loop. The optimization problems are posed as non-linear least squares and solved with a gradient-based algorithm within a trust-region scheme. The framework is applied across six cases: an industrial fin-plate process with quality issues, a new two-step route to achieve small radii (hydromechanical deep drawing followed by bulging-pressing), the same route on a complex geometry, a hydraulically expanded tube–tubesheet joint, an Airbus A380 door frame corner with springback and thinning-related fractures, and an exploration of residual stress relief in a deep drawn cup using eigenmode analysis. Across the cases, the approach shows that automatic optimization can identify relevant design and process parameters and, where appropriate, improve outcomes such as tool geometry and springback; for the door frame case, an optimized die shape improved springback. Finally, the framework is evaluated and guidelines are outlined for implementing such optimizations in practice.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents